Newer milestone thread

[petrw1]

Originally Posted by LaurV

I wonder which will come first:
- proving that the 47th known mersenne prime** is the 47th mersenne prime
- finding a 48th known mersenne prime

(**technically, the 46th known mersenne prime, in order of discovery, iirc. Or is it the 45th?).

Considering the former is likely at least 5 years away I hope we have found 48,49 and 50 by then.

[aketilander]

Originally Posted by petrw1

Considering the former is likely at least 5 years away I hope we have found 48,49 and 50 by then.

Well I am not too sure about that. We have not seen a new Mersenne prime in more than 3 years and our best estimate is that we will see one in about 6 years (2083 days):

Originally Posted by davieddy

2012-03-21 1.107
decrease = .024
days = 50 (Yep. Remembered the leap year)

Expected days to next prime = 50/.024 = 2083.

David

So I would be very surprised if we would find 3 Mersenne primes in 5 years or less. But it would be nice though if we did.

[bcp19]

Originally Posted by aketilander

Well I am not too sure about that. We have not seen a new Mersenne prime in more than 3 years and our best estimate is that we will see one in about 6 years (2083 days):



So I would be very surprised if we would find 3 Mersenne primes in 5 years or less. But it would be nice though if we did.

Where do you estimate the wavefront to be in 6 years? I know there are several people who will vociferously argue that there are no discernable patterns in predicting where the next prime may be, but people always see patterns even where there are none. So, I will make this disclaimer: I know this is pure speculation and there is no need for the usual denouncing by those who feel you need a PHD to participate here.

Now then, the 'perceived' pattern. If you look at the known Mersenne primes, you have groups of numbers that fit a sort of pattern. Almost double and more than double. If you make groups like this you have 4,a,4,a,4,m,2,m,1,a,5,m,3,a,3,a,1,a,3,a,1,m,4,m,2,m,1,a,1,a,8 which seems to have a pattern of sorts until the last group. Group them by more than double and it become 12, 2, 6, 11, 4, 2, 10 which appears more pattern-like. Again, this is pure speculation, but if the more than double pattern holds, then we could expect 1 of 2 things, another prime or 2 to be found within our current search area, or a huge dry spell.

[Dubslow]

Davieddy gets his data from mersenne.org/report_classic/, which I believe is based off a Poisson distribution.

[aketilander]

Originally Posted by bcp19

Where do you estimate the wavefront to be in 6 years?

Well, for each unfinish exponent you can estimate the probability of it beeing a Mersenne prime based on how much work has already been done on that exponent.

With the current speed of GIMPS we will have finished exponents with the probability of beeing prime summing up to 1 in the next 2083 days (6 years) so it could be argued that the best estimation we can give of the time to the next find of a Mersenne prime is 6 years, but the estimation is only based on probability it can be tomorrow or in 20 years, nobody knows.

So its like looking out of the window counting how many yellow cars you see a day (N) and use that as an estimate of how long it will take before you see the next one 24/N (in hours). Somtimes it takes longer time somtimes it takes shorter time.

(Sorry davieddy now I have messed it all up!!!)

[petrw1]

But I'm not holding my breath for the last. There were 12 under 40M last fall and still 9. Some details missing below to protect the innocent (not me).

Stage, %	age	to go	Completion	Next	        Updated	        Assigned
S2             	596	23	5/20/2012	4/28/2012	4/27/2012	9/9/2010
LL, 55.60%	1045	42	6/8/2012	4/26/2012	4/25/2012	6/17/2009
LL, 74.80%	801	-723	5/5/2010	6/5/2012	4/1/2012	2/16/2010
LL, 71.80%	668	49	6/15/2012	4/28/2012	4/27/2012	6/29/2010
LL, 70.90%	668	50	6/16/2012	4/28/2012	4/27/2012	6/29/2010
LL, 73.10%	668	47	6/13/2012	4/28/2012	4/27/2012	6/29/2010
LL, 72.40%	668	48	6/14/2012	4/28/2012	4/27/2012	6/29/2010
        	10	43	6/9/2012	5/2/2012	4/25/2012	4/17/2012
LL, 58.70%	1104	426	6/27/2013	4/26/2012	4/25/2012	4/19/2009

[Dubslow]

That S2 has been like that since at least the end of January, probably a lot longer. I'm thinking we should reawaken that thread.

[chalsall]

Originally Posted by Dubslow

That S2 has been like that since at least the end of January, probably a lot longer. I'm thinking we should reawaken that thread.

Oh, come on. Let's be honest.

Seven of those nine have been outstanding for more than two years.

GPU72 holds one. And it is receiving special attention.

Required deference to Cheesehead aside, I question why those who hold first time tests are allowed to hold them for years.

[Dubslow]

Originally Posted by chalsall

Oh, come on. Let's be honest.

Seven of those nine have been outstand for more than two years.

GPU72 holds one. And it is receiving special attention.

Required deference to Cheesehead aside, I question why those who hold first time tests are allowed to hold them for years.

I mostly agree, though we both know that the S2 is totally an accident (though as yet not brought to the attention of those who can fix it), as opposed to deliberate hoarding (I'd refrain from calling that malicious; "misguided", perhaps?).

[petrw1]

Originally Posted by chalsall

GPU72 holds one. And it is receiving special attention.

Required deference to Cheesehead aside, I question why those who hold first time tests are allowed to hold them for years.

GPUto72 does NOT have the S2 in reference.
Why do you say "special attention"?

I wonder if the "expiration routines" only focus on the current wave-fronts? Or can't be certain of the progress of V4 or Manual Assignments?
Case in point, there are 5 DCs in the 23M range that are:

1. Firstly, not even required
2. Over 2 years old
3. Showing now progress, mind you they are Manual Assignments
4. But are updating regularly

[chalsall]

Originally Posted by petrw1

GPUto72 does NOT have the S2 in reference.

I didn't say it did.

Originally Posted by petrw1

Why do you say "special attention"?

Because the candidates we are assigned below 45M receive special attention.